Gold Recovery Predictions at Zyfra IT Solution for Heavy Industries
The project assigned to me for my second integrated project at tenth sprint.
Throughout this sprint, I am striving to synthesize and employ all the insights and knowledge that I have accumulated throughout Sprint 6 - 9.
As a Data Scientist at Zyfra Company, I was entrusted to design a predictive model that capable to estimate the quantity of gold extractable from gold ore. The data related to both the extraction and purification processes of gold ore were provided as invaluable resources for training the model. The ultimate expectation of this model is to facilitate a more efficient production process.
Upon a deep analysis and construction of the model, the following encapsulates the project's summary:
In this project, I developed a model to forecast the amount of gold that can be extracted, based on the available features. Finally, I created three distinct models using Linear Regression, Random Forest, and KFold methodologies. A comprehensive examination of these three models revealed that the Random Forest model outperformed the others. When employing hyperparameters such as max_depth = 4, n_estimators = 50, and cv = 4, the model yielded an 8.04 sMAPE value on the test dataset.